Coding Agents

Coding Agents help developers generate, update, and enhance MuleSoft implementation code using natural language prompts. These agents streamline tasks such as API spec generation, DataWeave transformation scripting, and full-flow creation, with support for working from scratch, with repositories, or local project uploads.

Key Capabilities

• Prompt-Driven Code Generation: Turn plain English descriptions into Mule flows, components, and configuration code.
• Flexible Input Modes: Start with no repo or connect an existing repository, or upload a local MuleSoft project folder.
• Multi-tasking: These specialized handle everything from API spec creation to DataWeave scripting.
• Contextual Understanding: Generate output that aligns with your Mule/Java version, runtime config, and business rules.
• Interactive Feedback Loop: Review generated code, provide feedback, and approve the final version before applying.

---

Featured Coding Agents

1. API Spec Generator

Generates RAML 1.0 or OAS 3.0 specifications from natural language prompts.

• Supports three input modes: No Repository, With Repository, and Upload from Computer.
• In repo-based modes, the agent can analyze existing implementation code to generate or enhance the spec structure.
• Ideal for creating API specifications from scratch or updating specs based on existing MuleSoft projects.

2. Repository Coder

A multi-tasking AI agent designed to help you both build new MuleSoft implementation code from scratch and enhance existing projects.

• Supports all three input modes: No Repository, With Repository, and Upload from Computer.
• Capable of generating complete Mule flows, DataWeave transformations, connector configurations, and testable logic.
• Understands your existing repository and builds intelligently on top of it.
• Ideal for tasks like creating new flows, enhancing existing logic, upgrading connectors or runtimes, and more.

Additional Features (Optional):

• Run MUnit Tests: If required, check the MUnit run option to automatically execute tests after code generation.
• Diagram-Based Prompting: Optionally upload a sequence or flow diagram to help the agent structure your implementation.
• Runtime & VM Configuration: Select Mule/Java runtime versions and provide custom VM arguments, if needed, to match your environment.
• settings.xml Support: Upload a settings.xml file if you need to customize connector resolution or Maven settings.

This agent brings contextual awareness and flexibility, making it an ideal co-pilot for both greenfield development and iterative enhancements.

3. DataWeave Generator

Generates accurate and reusable DataWeave (DWL) scripts based on sample input-output payloads — no prompt required.

• Produces clean, optimized transformations without needing code context or instruction prompts.
• Just select the input format and provide matching input/output samples to generate DWL.
• Great for building new transformations or refining existing ones using real data.

Supported Input Formats:

• XML, JSON, CSV, YAML, TEXT, EDI, EDIFACT

Supported Input Modes:

• No Repository: Generate DWL purely from structured examples.
• With Repository: Reference files from an existing repo to guide generation.
• Upload from Computer: Import a MuleSoft project to add additional context.
• With Mapping Table: Upload a structured CSV mapping table to automatically generate a DataWeave script.

This agent is ideal for building robust data transformations fast, without needing to write DWL manually.

Best Practices

• Be specific in your prompt to get high-quality results (e.g., “Create a RAML spec for a Customer API with GET and POST”).
• Use the Notes field to add naming conventions, rules, or expected behaviors.